Statistical modelling and parametric optimization in document fragmentation

In recent days, most of the business enterprises and individuals are attracted towards cloud computing due to its cost efficiency and scalability. Though the cloud adoption is significant, its security has become nightmare due to its multi-tenancy property. Generally cloud service providers commit to ensure data reliability and security, but they may get depleted due to the rapid growth rate of cloud customers. To overcome the security issues and to protect the documents uploaded in cloud, cryptography is more used. Data security can further be improved with a technique called fragmentation which helps in outsourcing data partitions instead of entire document. The fragmentation becomes a difficult and time-consuming process when the size of document grows. In this paper, an efficient fragmentation process with virtualization is proposed. CPU cycles are efficiently used by the generation of VMs which reduce the time complexity of fragmentation process. The factors such as document size, processor capacity, storage capacity and number of VMs are taken into consideration to analyse their influence on the fragmentation time. Healthcare documents’ fragmentation process is conducted, and measured real-time values are analysed statistically. For experimentation purpose, a private cloud OpenStack on Oracle virtual box is used. Taguchi technique (L27 orthogonal array) is employed to find the optimum levels of the parameters on the fragmentation time, while analysis of variance is used to analyse the contribution of the parameters towards the performance of fragmentation process. Results reveal that the document size is the most dominant factor influencing the fragmentation time followed by processor speed. By parallelizing the fragmentation process with the help of multiple VMs, the time complexity of the process gets reduced.